Decisions made by surgeons, pathologists, and other physicians during cancer therapy depend largely on how the malignancy is classified, a process limited historically to histological criteria. Recent data suggest that the classification process can be improved significantly when the gene expression profile of the tumor is known. The most complete profiles have been obtained using microarray methods, which have the capabilities of monitoring the expression of tens of thousands of genes simultaneously. These analyses have identified mRNA transcripts that are most important for classifying a variety of tumors. Microarray techniques are not readily applicable to single cells, however. Data produced by analyses of whole tissues include values that reflect gene expression in non-transformed as well as transformed cells. Furthermore, tumors may contain more than one transformed cell type, a phenomenon that confounds this analysis and that may lead to errors in classification. Our desire to circumvent the limitations inherent in microarray analyses of whole tissues led us to seek alternate strategies for obtaining this information. We propose to build an instrument capable of monitoring simultaneously several mRNA transcripts in single cells of frozen sections. Its design makes use of well-known properties of fluorescence spectroscopy and should enable these data to be acquired during surgery. This will permit pathologists and surgeons to better assess the nature of the tumor at the time when decisions must made regarding the extent of the surgical procedure. It will also permit physicians to better tailor subsequent treatments that may involve radiation and chemotherapy, treatments that can have debilitating side effects.